arX

iv:a

stro

-ph/

9908

285v

1 2

6 A

ug 1

999

Mon. Not. R. Astron. Soc. 000, 1–14 () Printed 24 April 2018 (MN LaTEX style file v1.4)

Constraining the Star Formation Histories of Spiral Bulges.

R.N. Proctor, 1 A.E. Sansom,1 and I.N. Reid.21 Centre for Astrophysics, University of Central Lancashire, Preston, PR1 2HE, UK2 California Institute of Technology , 105-24, Pasadena, CA 91126,USA

ABSTRACT

Stellar populations in spiral bulges are investigated using the Lick system of spectralindices. Long-slit spectroscopic observations of line-strengths and kinematics madealong the minor axes of four spiral bulges are reported. Comparisons are made be-tween central line-strengths in spiral bulges and those in other morphological types(elliptical, spheroidal (Sph) and S0). The bulges investigated are found to have centralline-strengths comparable with those of single stellar populations of approximately so-lar abundance or above. Negative radial gradients are observed in line-strengths, sim-ilar to those exhibited by elliptical galaxies. The bulge data are also consistent withcorrelations between Mg2, Mg2 gradient and central velocity dispersion observed inelliptical galaxies. In contrast to elliptical galaxies, central line-strengths lie within theloci defining the range of <Fe> and Mg2 achieved by Worthey’s (1994) solar abun-dance ratio, single stellar populations (SSPs). The implication of solar abundanceratios indicates significant differences in the star formation histories of spiral bulgesand elliptical galaxies. A “single zone with in-fall” model of galactic chemical evolu-tion, using Worthey’s (1994) SSPs, is used to constrain the possible star formationhistories of our sample. We show that the <Fe>, Mg2 and Hβ line-strengths observedin these bulges cannot be reproduced using primordial collapse models of formationbut can be reproduced by models with extended in-fall of gas and star formation (2-17 Gyr) in the region modelled. One galaxy (NGC 5689) shows a central populationwith a luminosity weighted average age of ∼ 5 Gyr, supporting the idea of extendedstar formation. Kinematic substructure, possibly associated with a central spike inmetallicity, is observed at the centre of the Sa galaxy NGC 3623.

Key words: galaxies: abundances, galaxies: evolution, galaxies: formation, galaxies:spiral, galaxies: stellar content.

1 INTRODUCTION

The recently developed Lick system of spectral indices(Faber et al. 1985; Worthey et al. 1994) has provided pow-erful tools for the investigation of composite populations.Diagnostic plots, such as <Fe> vs Hβ and Mg2 vs <Fe> ,(where <Fe> is the average of Fe5270 and Fe5335 indices)break the age/metallicity degeneracy and have illustratedcharacteristics and trends in elliptical galaxies that constraintheir possible star formation histories (SFH). This papercompares the line-strengths in four spiral bulges with thoseobserved in other morphologies and places constraints onpossible SFHs in the sample. Comparison of the SFH of spi-ral bulges to that of elliptical galaxies is informative as thetwo morphologies show numerous similarities. Bulges exhibitsimilar surface brightness profiles and kinematic structuresto ellipticals and form an overlapping continuation of the el-liptical locus in the Fundamental Plane (Bender, Burstein &Faber 1993). Colours and colour gradients in the two typesof spheroid are also similar (Balcells & Peletier 1994). How-

ever, despite similarities in physical and photometric prop-erties, the two galaxy types are found in very different envi-ronments, with the majority of ellipticals located in clusters,while spiral galaxies are preferentially found in the field. An-other important factor in the environment of spiral bulgesis the presence of the disc, which severely limits possiblemerger histories.

Observations of elliptical galaxies have shown them topossess high central line-strengths. Gradients are observed inmetallicity sensitive features indicative of dissipation in theformation process. The Hβ index shows no indication of sys-tematic gradients in age (Carollo, Danziger & Buson 1993;Davies, Sadler & Peletier 1993; Fisher, Franx & Illingworth1996; Gorgas et al. 1997; Vazdekis et al. 1997). Modelling ofLick indices in elliptical galaxies, for comparison to observa-tion, has been carried out by a number of authors (Worthey,Dorman & Jones 1996; Greggio 1997; Vazdekis et al. 1997;Sansom & Proctor 1998). These studies have shown that the

c© RAS

2 R. Proctor et al.

high line-strengths found at the centre of elliptical galaxiescan only be achieved if the bulk of the observed populationforms from pre-enriched gas. Two possible processes havebeen proposed to produce this pre-enrichment. One possi-bility is an initial mass function (IMF) biased towards highmass stars in the early stages of galaxy formation. In thisway the first few generations of stars generate large quan-tities of metals (Gibson & Matteucci 1997; Vazdekis et al.1997). Alternatively, a period of star formation (SF), andthus interstellar medium (ISM) enrichment, with a SalpeterIMF prior to the main burst, can produce the necessarypre-enrichment. This delayed burst of SF may correspondto a merger/interaction model of elliptical galaxy formation(Sansom & Proctor 1998).

The second important feature of line-strengths in el-lipticals is their high central Mg2 with respect to <Fe>when compared to models assuming solar abundance ratios(Worthey, Faber & Gonzalez 1992; Davies et al. 1993; Gor-gas et al. 1997; Greggio 1997). This Mg2 excess is interpretedas indicating [Mg/Fe] greater than solar (Davies et al. 1993).High [Mg/Fe] can be achieved by star formation times of <∼1 Gyr, in accordance with the short time-scales of gas in-flow in both merger/interaction (Barnes & Hernquist 1996)and primordial collapse models of elliptical galaxy formation(Theis, Burkert & Hensler, 1992). Short bursts of star forma-tion ensure that the metal enrichment of the ISM from whichthe bulk of the population forms, is mainly produced in theType II supernovae that are the main sites for productionof Mg. The biased IMF proposal also naturally produces anexcess of Type II supernovae (and consequently Mg2) due toan over-production of high mass stars. While both mergerand primordial collapse (with biased IMF) models can suc-cessfully reproduce the indices in the Mg2 vs <Fe> plane,the high central Hβ values of many elliptical galaxies are achallenge to primordial collapse models. In contrast, mergermodels clearly predict the existence of younger populationsin these objects.

Observations of other galaxy morphologies, such as S0s(Sil’chenko 1993; Fisher et al. 1996) and Sphs (Gorgas etal. 1997), show differences between central line-strengths invarious galaxy types in the <Fe> vs Mg2 plane. S0s tendto exhibit weaker Mg2 for a given <Fe> than do Es whilestill showing an excess with respect to values for solar abun-dance ratio SSPs. Gorgas et al. (1997) showed that Sphsroughly follow solar abundance ratio SSP loci in the <Fe>vs Mg2 diagnostic plot. Low luminosity spiral bulges ob-served by Jablonka, Martin & Arimoto (1996) also span theregion of the Mg2 vs <Fe> plot occupied by Worthey’s so-lar abundance ratio SSPs (Worthey 1998), while some highluminosity bulges in their sample exhibit greater than solar[Mg/Fe].

Observations of individual stars in the bulge of our ownGalaxy, while hampered by obscuration, have provided in-teresting insights into the population present. Minniti et al.(1992) report a broad range of metallicities (−2 <[Fe/H]<+1) in K giants in the Galactic bulge and indicate the pres-ence of a metallicity gradient. Observations of resolved Mgiants in the centre of the bulge of our own Galaxy indicateapproximately solar abundances and also indicate a gradientin metallicity (Frogel 1998). Recent episodes of star forma-tion at the centre of the bulge, with several bursts havingoccurred over the past few 100 Myr, have also been sug-

gested (Blum, Sellgren & Depoy 1996; Frogel 1998). Theseobservations suggest that the population of the bulge of ourown Galaxy has a complex and extended star formation his-tory.

Our long-slit observations of spiral bulges allow compar-ison of central line-strengths and gradients to those in othergalaxy types and constrain possible SFH in the bulges. Weprobe whether the similarities between bulges and ellipticalgalaxies in morphological and photometric properties are theresult of similar formation processes, or are rather the resultof the evolution of differing histories to similar present-dayforms. In Section 2 the observations, data reduction and er-ror estimates are outlined. Results of our analysis and com-parison with other Hubble types are reported in section 3.Comparison to models are detailed in section 4. Our conclu-sions and the direction of future work are outlined in section5.

2 OBSERVATIONS & DATA REDUCTIONS.

2.1 Observations.

Using the Palomar 5.08 meter telescope on 1995 March 31,long-slit spectroscopy was carried out along the minor axesof the bulges of 4, approximately edge-on, spiral galaxies(table 1 and fig. 1). We chose to observe edge-on systemsalong their minor axes as disc obscuration of bulge light isthus concentrated on one side of the bulge. A fifth galaxy(NGC 4013) was also observed but appeared to have a brightforeground star slightly offset from the centre of the galaxy.Consequently, this galaxy was excluded from data reduc-tions. The Double Spectrograph was used (Oke & Gunn1982) with a 5700 A dichroic filter and Texas InstrumentsCCDs on both red and blue arms. Wavelength coverage inthe blue was 4560 - 5430 A at a resolution of 1.1 A per pixel.In the near infra-red the range 8235 - 8860 A was observedat a resolution of 0.8 A per pixel. The blue wavelength rangewas chosen to include the Hβ, Mg and Fe absorption featurescalibrated by Faber et al. (1985) and extensively observed inelliptical galaxies (Davies et al. 1993; Hes & Peletier 1993;Fisher et al. 1996; Gorgas et al. 1997; Vazdekis et al. 1997).The red wavelength range spans the CaII IR triplet and wasintended for the study of both indices and kinematics. Spa-tial resolutions of the CCDs were 0.78 and 0.58 arcsec perpixel on blue and red arm respectively. Seeing was ∼ 1.5arcsec and a 1 arcsec slit was used. Flux and velocity stan-dard stars were also observed. Object frames were alternatedwith arc lamp exposures to allow for accurate wavelengthcalibrations. Frames of a continuum source (tungsten lamp)were also obtained for the purpose of flat-fielding. Galaxyexposure times of 5900, 7200, 4200 and 1200 seconds wereobtained for NGC 2654, NGC 3623, NGC 4565 and NGC5689 respectively. A maximum single exposure time of 2400seconds was adopted to facilitate cosmic ray removal.

2.2 Data Reduction.

Unless otherwise stated, data reduction was carried out us-ing the FIGARO package of Starlink software. A bias levelfrom the over-scan region was subtracted from each frame.

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 3

Table 1. Data for four spiral bulges observed with the Palomar 5m telescope. Hubble Type and B magnitudes are from de Vaucouleurset al. (1991) (hereafter RC3). Inclinations are from Guthrie (1992). Distances are from Tully (1988) assuming Ho=75 kms−1Mpc−1.Recession velocities given are from RC3 (HI data) and this work (optical data). The velocity dispersions given (σV ) are the averagesacross all data points. This is the value used for calculation of broadening required to the Lick system (see section 2.3).

Galaxy Hubble Inclination Distance B Recession Velocity σV ScaleType (deg) (Mpc) (mag) RC3/This work (kms−1) (pc/arcsec)

(kms−1)

NGC 2654 Sab 90 23.3 12.8 1341/1320 136 110NGC 3623 Sa 79 7.3 9.6 807/809 152 35NGC 4565 Sb 90 9.7 10.3 1227/1235 155 46NGC 5689 S0/a 78 35.6 12.7 2160/2221 149 168

N

E

Figure 1. Digital Sky Survey images of galaxies with slit positions along minor axes. The slit length is 2 arcmin in all images. Slitposition was determined by assuming the slit passes through the telescope pointing position along the minor axis. The location of thebulge within the slit was determined by ensuring that the peak of the luminosity profile on the slit matched the position of the maximumcount rate on the CCD. Upper left: NGC 2654. Note that the slit appears off-centre by approximately 5 arcsec. Upper right: NGC

3623. Sky is not reached on either side of the galaxy. However, sky was estimated from extreme edges of slit (see section 2.2). Lower

left: NGC 4565. Sky is not reached on the side of the galaxy most affected by the disc. However, a reasonable sky estimate is achievedon the side least affected by disc. Lower right: NGC 5689. Note the slit appears off-centre by approximately 5 arcsec.

Due to the presence of localised features in the flat-fieldframes, each object was divided by the normalised, averageflat-field. This biases the data with the spectral response ofthe tungsten lamp. However, this is smooth and is removedduring flux calibration. Cosmic rays were removed by in-terpolation between adjacent pixels. Wavelength calibrationwas achieved by interpolating between arc lamp exposures

bracketing galaxy frames and was accurate to ∼ 0.1 A. Spec-tra were then flux calibrated and extinction corrected. ForNGC 2654, NGC 3623 and NGC 5689 sky subtraction wascarried out by interpolation across the galaxy. For NGC 3623sky was not fully attained on either side of the galaxy. Thisresults in a small contribution of disc light to the sky es-timate. The error associated with this sky was estimated

c© RAS, MNRAS 000, 1–14

4 R. Proctor et al.

from the 1 sigma variation in index values when sky esti-mates from successive galaxy exposures (NGC 2654, NGC3623 and NGC 4565) were used. For NGC 4565 sky wasonly reached along the slit on one side of the bulge. In thiscase the sky was estimated from just this side of the galaxy.Variation of the sky across the slit was shown to be less than10% of sky level for frames clearly reaching sky on both sides.Finally, galaxy frames were shifted and co-added to form asingle frame for each galaxy.

2.3 Measurement of Kinematics and Indices.

Kinematics were measured in order to correct our data tothe dispersion of the Lick system. Measurements of kinemat-ics were carried out on both red and blue data using bothcross-correlation and Fourier quotient techniques within theIRAF software package. Results of the two techniques werein agreement to within 5 kms−1. However, the Fourier quo-tient technique was used as the statistical errors associatedwith cross-correlation were substantially larger due to theinsensitivity of this method at the low velocity dispersionstypical of spiral bulges. As velocity dispersions in our sam-ple are all less than the dispersion required for calibration tothe Lick system, the additional broadening required ( σB)can be calculated (assuming Gaussian profiles) by;

σ2

B = σ2

L - σ2

I - σ2

V

The Lick resolution (σL) is approximately constant at 3.57Aover the 4500-5500 A range covered by our observations(Worthey & Ottaviani 1997). Instrumental broadening (σI)was estimated from arc lines and found to be 1.12 A. Veloc-ity dispersion broadening of the galaxy (σV ) was estimatedas the average over the whole galaxy in the blue (table 1).

After broadening the spectra to the Lick resolution, theindices were evaluated by our own code, using the bandwavelength range definitions supplied by G. Worthey on hishome page. Worthey also kindly provides data with whichour code could be tested. Differences between the Wortheyderived indices and our own for the provided spectra were≤ 0.03 A for the line features and ≤ 0.002 mag for the Mg1and Mg2 molecular bands. These discrepancies are smallerthan differences caused by re-calibration of Worthey’s datato our wavelength resolution and are probably the result ofthe effects of differences in the handling of partial bins. Theyare also significantly smaller than the total errors (section2.4). A more important source of uncertainty in our data isthe conversion to the Lick system (which was not flux cali-brated). One of the observed velocity standards, HD 139669is also one of the Lick calibration stars. Indices derived fromour observations of this star and the data of Jones (1996)(data available on AAS CD-ROM Series Vol. VII) were com-pared to the Gorgas et al. (1993) published data. Differencesbetween indices derived from our data and those of Jonesare typically within errors. However, there is a large dis-crepancy between our measurement and the published Lickdata for the Hβ and Fe5015 indices (also reproduced in thedata of Jones) which is probably caused by unpredictablewavelength shifts in the Lick IDS spectra (private commu-nication with J. Gorgas). Differences between our measuredindices and the published Lick data for this star are givenin table 2. HD 139669 (HR 5826) also lies well away fromthe mean in fig.10 of Gorgas et al. (1993). Therefore, this

star is not suitable for estimation of correction factors to theLick system. Lack of sufficient data to fully calibrate to theLick system represents a limitation on the Palomar data set.However, corrections to the narrow line features discussedhere (Fe4668, Hβ, Fe5270 and Fe5335) have been shown tobe small (Gorgas et al. 1997; Vazdekis et al. 1997; Worthey& Ottaviani 1997). For Jones’ (1996) observations correctionto the flux calibration sensitive Mg2 index was found to besmall (0.01 mag increase in the measured Mg2 index). Thegood agreement between our measurement of the Mg2 indexof HD 139669 and that of Jones suggests that our correctionfactor to the Lick system should be of similar magnitude (∼+0.01 mag), although we have not applied this correction.

Initial measurements of indices showed unusual gradi-ent inversions in the CaII, Mg1 and Mg2 indices at the cen-tres of the bulges (defined as the peak in brightness). Thesewere found to be the result of focus variation caused bydistortion of the CCD surface. This effect redistributes fluxspatially between bins, introducing false continuum varia-tions. CaII and molecular band indices are sensitive to thistype of distortion due to the large separation of their side-bands. The problem was most pronounced in the red (thisCCD has subsequently been replaced at Palomar). As a re-sult, line-strength measurements were not carried out onthe red arm data. Kinematic analysis were still possible asthe continuum shape is removed prior to line-broadeningmeasurement. Stellar data from the blue arm had a smoothpoint source FWHM variation of <20% across the wave-length range. Even so, features are still introduced in theMg1 and Mg2 indices in the centre of the galaxies whereluminosity gradients are greatest. A minimum binning of ∼1.5 arcsec has minimised this effect and matched the see-ing conditions. Central values for purposes of modelling andcomparison to ellipticals were calculated using the central 3arcsec (see section 3.2).

2.4 Estimation of Errors.

Detailed analysis of errors affecting NGC 3623 are shownin table 2 for illustration. For all galaxies, binning of datawas selected to maintain an approximately constant Poissonerror outside the central regions. For NGC 2654, NGC 4565and NGC 5689, sky errors were evaluated by varying theregion of sky used for sky estimation. The 1 σ variation inestimated index value is taken as the error. For NGC 3623,which failed to reach sky on both sides of the galaxy, skyerror was estimated as the 1 sigma variation in indices de-rived utilising sky estimates from NGC 3623 and the brack-eting observations of NGC 2654 and NGC 4565. Variationof both velocity dispersion and recession velocity about theassumed average values (table 1) were both ∼ ± 35 kms−1.These variations are significantly larger than errors associ-ated with wavelength calibration and statistical errors fromthe Fourier quotient technique (∼ 9 kms−1). Consequently,velocity dispersion errors were estimated as the variation inindices caused by increasing the galaxy velocity dispersionassumed in calculation of broadening to the Lick system by35 kms−1 . Index errors due to recession velocity uncertaintywere estimated as the variation in indices caused by increas-ing the red-shift estimate by 35 kms−1.

The magnitude of the sky error varied between galax-ies. Consequently, unless otherwise stated, figures and tables

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 5

Table 2. Error table for NGC 3623. Above: Index values for the central 3 arcsec and the associated Poisson noise are given. Sky errorsare derived as detailed in section 2.4. Calculations of both velocity dispersion and recession velocity errors (σ and V respectively) assumeapproximate velocity uncertainties of ± 35 kms−1. Total errors are Poisson, sky, σ and V errors added in quadrature. Comparison toHD 139669 represents published Lick values minus values obtained from our observation of this star. These offsets are not included inthe calculation of total error due to their uncertainty. Below: Error table for outer regions (∼ 8 arcsec from centre) of NGC 3623.

CENTRAL VALUES

Index Index Poisson Sky σ V Total ComparisonValue Noise Error Error Error Error to HD 139669

Fe4668 8.64 A 0.07 0.05 0.05 0.06 0.12 0.15Hβ 1.29 A 0.03 <0.01 0.03 0.02 0.05 0.60

Fe5015 4.91 A 0.07 0.07 0.23 0.04 0.25 -0.93Mgb 4.51 A 0.04 0.02 0.09 0.02 0.10 -0.03

Fe5270 3.31 A 0.04 0.01 0.10 0.01 0.11 -0.47Fe5335 3.07 A 0.05 <0.01 0.17 0.02 0.18 0.14Mg1 0.133 mag 0.002 <0.001 0.001 < 0.001 0.002 0.014Mg2 0.279 mag 0.002 <0.001 0.001 0.001 0.002 0.016

OUTER VALUES

Fe4668 6.88 A 0.12 0.28 0.07 0.05 0.32 0.15Hβ 1.22 A 0.06 0.01 0.01 0.01 0.06 0.60

Fe5015 3.47 A 0.12 0.47 0.21 0.11 0.54 -0.93Mgb 4.12 A 0.07 0.12 0.07 0.01 0.16 -0.03

Fe5270 2.79 A 0.08 0.03 0.09 0.01 0.12 -0.47Fe5335 2.59 A 0.09 0.02 0.13 0.01 0.16 0.14Mg1 0.107 mag 0.003 0.001 0.001 < 0.001 0.003 0.014Mg2 0.247 mag 0.004 0.003 0.001 0.001 0.005 0.016

show combined Poisson and sky errors. Velocity dispersionand recession velocity errors for other galaxies were simi-lar in magnitude to those in NGC 3623. In the centres ofthe bulges Poisson noise and velocity dispersion errors dom-inate. While for most cases sky error dominates in the outerregions, total errors in Fe5270 and Fe5335 continue to havesignificant contributions from velocity dispersion error. Gra-dient errors given in table 3 were estimated as the statisticalerrors associated with linear fitting.

3 RESULTS.

3.1 Kinematics.

Results of recession velocity and velocity dispersion analy-sis are shown in fig. 2. Galaxy centres are defined as theposition of the luminosity peak along the slit. Reasonableagreement exists between results for red and blue data withaverage values for both recession velocity and velocity dis-persion differing by less than 20 kms−1. However, systematicdifferences are present in the centres of NGC 3623 and NGC4565. Central recession velocities derived from the red dataare highly susceptible to systematic error due to focus varia-tion (section 2.3). Consequently, analysis of indices has beenbased on the blue recession velocity data only. In NGC 4565(a known LINER, Keel, 1983), the maximum in the blue ve-locity dispersion profile is offset by ∼ 5 arcsec with respectto both the luminosity and metallicity peaks (see figs 2 and3). This would appear not to be an internal extinction effectas metallicity continues to rise with luminosity to the centre.Photometric observations in the near-infrared of NGC 4565(Rice et al. 1996), rather than indicating reddening by dust

in the centre, show a flattening of the colour gradient that isevident along the minor axis at distances greater than 5 arc-sec (Rice et al. 1996; fig. 2a). The change in colour gradientis coincident with the peak in blue velocity dispersion datain fig. 2 and and the edge of the broad metallicity peak infig. 3. The central region of radius 5 arcsec is also the emis-sion region of NGC 4565. Strong [OIII] 5007 A and 4959 A

emission lines are present in this region affecting the Fe5015index, while the Hβ index exhibits features due to emissionline filling (fig. 3). Thus, the data suggest that the core ofNGC 4565 contains a blue, metal rich, kinematically coldsub-population perhaps associated with ongoing star forma-tion. The presence of such a population would also causea displacement between red and blue luminosity peaks asobserved.

A dip in velocity dispersion is seen in the centre of NGC3623 and is reproduced in both red and blue data sets. Suchvelocity dispersion profiles have been observed along the ma-jor axis of spiral galaxies (e.g. Bertola et al. 1996; Bottema &Gerriston 1997). Kinematic models that reproduce such dipsin velocity dispersion include the presence of a kinematicallydistinct component within the bulge such as an isothermalnuclear core (Bottema & Gerritsen 1997), a disc (Bertolaet al. 1996) or a bar (Friedli 1996). Chemo-dynamical mod-elling of the effect of the presence of a bar on disk galaxies(Friedli 1998) has shown that metallicity gradients are re-duced in the plane of the bar around and outside the regionof the co-rotation radius, while steeper gradients can occurat radii well within co-rotation. A flattening of the metallic-ity gradients is also induced along the minor axis. The gra-dients in metallicity sensitive indices along the minor axis ofNGC 3623 (figs 3), which flatten rapidly away from the cen-tre, are consistent with such models. Friedli’s (1994) models

c© RAS, MNRAS 000, 1–14

6 R. Proctor et al.

Figure 2. Spatially resolved kinematics along the minor axes of 4 spiral bulges. Recession velocities and velocity dispersions fromboth the blue and red Palomar data are shown. Both sets of data were derived by the Fourier quotient technique. Error bars representstatistical errors of this technique and do not include systematic errors.

also predict inflow of gas within the co-rotation radius re-sulting in a central burst of metal rich star formation. Thesharp metallicity peak, evident in both Fe and Mg indices infig 3, may be the result of this process. Photometric obser-vations by Burkhead et al. 1981 and recent Hubble archiveimages of NGC 3623 show boxy central isophotes which mayalso be indicative of the presence of a bar. Alternatively, thekinematic feature may be the result of an embedded stellardisc. If this is the case, either the disc has a scale-heightof ∼ 150 pc (major axis disc) or the disc rotates along theminor axis with a radius of ∼ 150 pc. However, there isno evidence of the kinematic signature of a minor axis discon this scale in the recession velocity measurements (fig. 2).On balance it would seem likely that the centre of this spiralbulge harbours a small scale bar. (Note also that RC3 givesthe classification of NGC 3623 as SXT1. This indicates a”mixed” bar/no bar type).

As all the indices described here are from the blue data,and the red data suffer degradation by focus variation, wehave used the blue velocity dispersion and recession velocityvalues in our analysis of the indices. Average values acrossthe galaxies were used for both recession velocity and veloc-ity dispersion in recognition of the systematic uncertainties.The resultant recession velocities are in good agreement withvalues given in RC3 (table 1).

3.2 Lick Indices.

In this section we present the results of index evaluationsand plot the data on several diagnostic plots for compar-ison to models of SSPs and observations of other galacticmorphologies. Index measurements across the 4 bulges areshown in table 3 and fig. 3. Negative radial gradients canbe seen in all the metallicity sensitive indices with the pos-sible exception of Fe5015, which is strongly affected by the[OIII] 5007 A emission line. Mgb is affected by emission ofthe [NI] 5199 A doublet in one of the side-bands (Goud-frooij & Emsellem 1996). However, the enhancement of thisindex caused by emission appears to be small in these datafor all galaxies, with the possible exception of the LINERNGC 4565. The age sensitive Hβ index can also suffer fromemission line filling, particularly near the centres of bulges.This is most clearly seen in NGC 4565 (fig. 3). Dips in Hβ

at the centre of this galaxy are associated with two separateemission regions and are reflected by features in the Fe5015index. The dip in Hβ in the central region of NGC 3623 mayalso indicate weak emission. However, this galaxy exhibitssharp central peaks in metallicity sensitive features whichcoincide with a dip in the velocity dispersion (see section3.1). The combined effects of the the influence of the metal-licity peak and emission on the Hβ index in the centre ofthis galaxy makes interpretation difficult. The symmetry ofindices about galaxy centres indicate that only in NGC 4565is disc obscuration significant.

Fig. 4 shows central values of Mg2 vs <Fe> for our fourspiral bulges. Also plotted are central index values of ellip-tical (E) and Sph galaxies (Gorgas et al. 1997), S0 galaxies

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 7

Figure 3. Indices across the minor axes of four spiral bulges. Radial gradients can be seen in most metallicity sensitive features. Thecentral dips in Hβ in NGC 4565 are the result of emission. Fe5015 is also emission affected. The errors bars shown include Poisson noiseand sky error only. Bin sizes were selected so as to maintain approximately constant Poisson error values beyond the central pixels. Theincreasing influence of the sky error at large radii can be seen. Note the sharp central peak in metallicity sensitive indices in NGC 3623.This peak coincides with a dip in central velocity dispersion.

c© RAS, MNRAS 000, 1–14

8 R. Proctor et al.

Table 3. Top: Central 3 arcsec index values (except Hβ for NGC 4565 where the average of outer values is used to avoid emission: seesection 3.2). Errors (given in brackets) are Poisson noise and sky errors combined. Middle: Index gradients calculated from all availabledata points. For NGC 4565, errors are statistical errors (associated with linear fitting technique) combined with uncertainty in position ofthe galaxy centre. Positional uncertainties are small in other galaxies, consequently, for these, quoted errors are purely statistical errors.Bottom: Central velocity dispersions. Quoted errors are statistical errors from the Fourier quotient technique. 1 Peak value 175 kms−1

at 5 arcsec off-centre. 2 Peak value 190 kms−1 at 5 arcsec off-centre.

Central Index Values

NGC 2654 NGC 3623 NGC 4565 NGC 5689

Fe4668 (A) 7.33(0.25) 8.64(0.08) 8.94(0.17) 8.72(0.26)Hβ (A) 1.47(0.09) 1.29(0.03) 1.49(0.08) 1.98(0.11)

Fe5015 (A) 4.69(0.27) 4.91(0.10) 2.49(0.16) 5.74(0.24)Mgb (A) 4.13(0.11) 4.51(0.04) 4.67(0.08) 4.03(0.13)

Fe5270 (A) 3.00(0.12) 3.31(0.04) 3.32(0.09) 3.00(0.14)Fe5335 (A) 2.67(0.14) 3.07(0.05) 3.14(0.10) 2.62(0.16)Mg1 (mag) 0.105(0.005) 0.133(0.002) 0.139(0.004 ) 0.095(0.006)Mg2 (mag) 0.244(0.007) 0.279(0.002) 0.293(0.005 ) 0.237(0.008)

Index Gradients (δ Index/δlog(radius))

<Fe> -0.60(0.44) -0.44(0.05) -0.58(0.10) -0.55(0.19)Fe4668 -0.61(0.98) -1.89(0.14) -2.20(0.33) -3.28(0.60)Mg2 -0.022(0.016) -0.032(0.003) -0.061(0.008) -0.027(0.012)Hβ -0.35(0.21) -0.06(0.04) 0.52(0.11) 0.25(0.20)

Central Velocity Dispersions

σ0 (kms−1) 145(7) 1451(7) 1622(7) 147(8)

Figure 4. Mg2 vs <Fe> for elliptical and Sph galaxies (Gorgas et al. 1997) and S0 galaxies (Fisher et al. 1996). Results for the 4 spiralbulges presented here and the bulge of the Sa galaxy M104 (Hes & Peletier 1993) are also shown. The dotted lines indicate the extremesof Worthey’s (1994) SSPs and also the solar metallicity locus (Z=0.0). The solid lines indicate the range of index values achieved byprimordial collapse models with a range of star formation efficiencies from 0.2 to 4.0 Gyr−1 (described in section 4). Dashed lines indicateindex values achieved by a range of extended inflow models with star formation efficiencies of 4.O Gyr−1 over 4 to 6 Gyr. Ages of 5, 8,12 and 17 Gyr are shown. All models plotted assume solar abundance ratios.

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 9

(Fisher et al. 1996) and the bulge of M104 (Hes & Peletier1993). This diagram illustrates trends in the relative abun-dance and abundance ratio of different morphological types.Dotted lines indicate the extent of the space occupied byWorthey’s solar abundance ratio SSPs and the solar metal-licity locus. The high central strengths of the <Fe> indexfound in E and SO galaxies are also apparent in spiral bulges,with bulge indices comparable to those of SSPs with so-lar metallicity or above. Fig. 4 also demonstrates that thecores of Es exhibit Mg2 values in excess of those possible forSSPs with Salpeter IMF and solar abundance ratios. Thisis the well known Mg excess in elliptical galaxies discussedin section 1. S0 galaxies lie intermediate between Es andthe SSPs. The position of the four spiral bulges is consistentwith solar abundance ratios. This is in agreement with theresult of Sil’chenko (1993) and Worthey’s (1998) interpre-tation of the Jablonka et al. (1996) data for bulges. Outerregions of the bulges are also consistent with solar abun-dance ratio SSPs (Sansom, Proctor & Reid 1998). Analysisof the flux calibrated data of Jones (Worthey & Ottaviani1997) indicated additive corrections of +0.01 mag and -0.06A for calibration of Mg2 and <Fe> features respectively tothe Lick system. While reasonable values for this correctionfor the Palomar data may move some of the spiral bulgedata points just outside the region of solar ratio SSPs, theconclusion that Mg is significantly under-abundant in spiralbulges, when compared to ellipticals, remains unchanged.The bulge of M104 (Hes & Peletier 1993) does not appearto follow the trend suggested by our sample (fig. 4), but in-stead falls amongst the elliptical galaxies. Therefore, thesedata support the suggestion by Hes and Peletier, that thebulge of M104 (classified as S0/a in RC3) more closely re-sembles an elliptical galaxy than a normal spiral bulge.

In fig. 5 we plot <Fe> against Hβ. This diagram pro-vides evidence relating to the relative ages of stellar popu-lations. The correction factors to transform flux calibratedHβ data to the Lick system have been shown to be small(Worthey & Ottaviani 1997). Index values from our dataare from the central 3 arcsec, except for NGC 4565 wherethe average outer Hβ value was used (omitting central valueswhere this index is emission affected). We therefore assumeno significant gradient in Hβ for NGC 4565. This is consis-tent with studies of elliptical galaxies (Carollo et al. 1993;Davies et al. 1993; Fisher et al. 1995; Fisher et al. 1996;Gorgas et al. 1997; Vazdekis et al. 1997) that show no de-tectable age gradients. The possible presence of emission inthe bulges creates uncertainty in the reliability of our Hβ

determination, making interpretation of this index difficult.However, it should be noted that NGC 5689 (with Hβ ∼

2.0 A) shows no sign of central emission which, in any case,would only increase the estimate of Hβ for the underlyingpopulation. Thus, the strong Hβ in this galaxy is compellingevidence of an intermediate age (≤ 5 Gyr) population.

Fig. 6 locates the spiral bulges in the Fe4668 vs Hβ

plane. The Fe4668 index is extremely metallicity sensitive(Worthey 1994). The correction factor to transform Fe4668to the Lick system has been shown to be small (Worthey &Ottaviani 1997). Comparison of figs 5 and 6 illustrates thathigher metallicities are suggested by Fe4668 than by <Fe>.As the Fe4668 index is highly carbon sensitive (Tripicco &

Bell 1995), this may indicate either an abundance ratio dif-ference or a problem with the calibration of Fe4668. Themain source of carbon enrichment of the ISM is still contro-versial. Gustafsson et al. (1999) report falling [C/Fe] withincreasing [Fe/H] in the solar neighbourhood. They suggestthat high mass stars are the major contributors of C to theISM. However, both intermediate and low mass stars arealso known to produce C. Consequently, the interpretationof a C over-abundance is difficult. It must also be noted thatat high metallicities the value of this index changes rapidlywith metallicity. It is possible that the seeming Fe4668 ex-cess is in fact the result of the lack of calibrated SSP data atmetallicities greater than [Fe/H] = 0.5. Caution is advisedin interpreting this index at high metallicities.

3.3 Gradients in Indices and Correlations.

Gradients were estimated by least squares fitting for the<Fe>, Fe4668, Mg2 and Hβ indices plotted versus log radiusfrom the galaxy centre (fig. 7). Data from both sides of thegalaxies were included in these fits. The gradients obtainedare given in table 3. Central galaxy regions are marginallyaffected by seeing. However, central index values were in-cluded in the gradient estimates due to the limited numberof data points. As gradients are unaffected by a constantoff-set to convert to the Lick system, a direct comparisonwith elliptical galaxies can be made. Gradients, central Mg2values and central velocity dispersions (σ0; table 3) of thebulges are consistent with the correlations reported by Car-ollo et al. (1993) in elliptical galaxies. The bulges show Mg2gradients of similar magnitude to those in elliptical galaxiespossessing the same central velocity dispersions and centralMg2 values. The gradients are also similar in magnitude tothose reported in Sansom, Peace & Dodd (1994) for 2 bulges(NGC 3190 and NGC 1023). The central velocity disper-sions and the central index values of our Palomar sample,indicating solar abundance ratios, are also consistent withthe pattern suggested by Worthey (1998) that all spheroidswith velocity dispersions less than 225 kms−1 possess solar[Mg/Fe].

4 MODELLING CENTRAL

LINE-STRENGTHS.

In order to investigate possible SFHs, we use our galac-tic chemical evolution (GCE) code detailed in Sansom &Proctor (1998). Briefly, the model calculates star formationrate (SFR) and the metallicity of the gas at each time-step throughout the history of the region being modelled.As all stars formed in each time step are of the same ageand metallicity, they constitute an SSP. Composite indicesare then calculated as the luminosity weighted sum of in-dices from interpolations between tabulated SSP values ofWorthey (1994) which cover populations of age 1.5 to 17Gyr. The code currently models a single zone allowing forgas in-fall. The rate of gas inflow can be varied and can beeither primordial or enriched to the same level as the gas al-ready in the region (simulating inflow from a neighbouringregion with similar SFH). SFR is assumed to be propor-tional to some power (α) of the gas mass density (ρ); SFR= Cρα, where C is the star formation efficiency (Gyr−1) if

c© RAS, MNRAS 000, 1–14

10 R. Proctor et al.

Figure 5. <Fe> vs Hβ for E, S0 and Sph galaxies as well as M104 and the 4 spiral bulges presented here. Data sources are as in fig.4.Dotted lines indicate Worthey’s SSP, with values of Z = [Fe/H]; -0.5, -0.25, 0.0, +0.25, +0.5 and ages; 1.5, 2, 3, 5, 8, 12 and 17 Gyr. Itmust be noted that due to the effects of emission on the central Hβ index for NGC 4565 and M104, the plotted Hβ values are averagedouter values. Solid and dashed lines are as described in fig 4 (see section 4).

α=1. Kennicutt (1989) shows that the value of α in galacticdiscs lies between 1 and 2. In all models presented here weassume α=1 and inflow is enriched. The differences in in-dices caused by assuming α = 2 were shown to be small andmake no significant difference to the conclusions for the mod-els presented here. The code permits the modification of thestar formation efficiency and inflow rates at two points in thehistory of the galaxy. In this way various formation scenar-ios can be modelled. The star formation efficiency modelledranges between values for typical Sb galaxies (C = 0.2) up tothose for star-burst galaxy types (C = 4). In order to followthe enrichment of the gas by feedback mechanisms, such asSNII, SNIa and mass-loss from intermediate mass stars, thetotal metal content of the gas is traced in a self consistentmanner. Data for SNII (high mass stars) were taken fromWoosley &Weaver (1995) and Maeder (1992). SNIa data arefrom Nomoto, Thielmann & Yokoi (1984). For intermediatemass stars Renzini & Voli (1981) data were used. The codealso uses the SSPs of Weiss, Peletier & Matteucci (1995) toestimate the effects of non-solar abundance ratios on <Fe>and Mg2 predictions. Currently, our GCE code outputs esti-mates of 21 Lick indices for the population being modelled.

4.1 Primordial Collapse.

Primordial collapse represents the isolated collapse of a pri-mordial gas cloud. The cloud commences SF when the localdensity rises above some critical level. This mechanism hasbeen proposed for the formation of elliptical galaxies. Dy-namical models of elliptical galaxy formation by primordial

collapse (Carlberg 1984) show that the period of gas inflowis < 1 Gyr. The Chemo-dynamical models of of Theis etal. (1992) also show the main burst of star formation to becomplete in ≤1 Gyr. Thus, these models predict a SFH com-prising a single burst of SF with a large majority of starsformed within the first 1 Gyr. This naturally leads to theprediction of [Mg/Fe] > solar, observed in the centres of el-liptical galaxies, as there is insufficient time for SN1a to con-tribute large quantities of Fe to the ISM before the bulk ofSF is complete. These models also successfully predict manyphotometric properties such as the de Vaucouleurs r1/4 pro-file and the presence of colour (metallicity) gradients. How-ever, several authors (Worthey et al. 1996; Greggio 1997;Vazdekis et al. 1997; Sansom & Proctor 1998) have shownthat primordial collapse models fail to reproduce the strongcentral, metallicity sensitive, line-strengths found in ellipti-cal galaxies, under the assumption of a constant, SalpeterIMF. This is the result of the high number of low metal-licity stars produced in this scenario. We find this to betrue also for the centres of spiral bulges which exhibit sim-ilar metallicities as the centres of elliptical galaxies (fig. 4).To demonstrate this point we have calculated indices for arange of primordial collapse models. Parameters within themodels were set to maximise predicted central metallicitieswhile keeping models realistic. Thus, as in-fall of primordialgas into the modelled region would dilute the metals con-tent of the star forming ISM, we assumed enriched inflow.The gas flow rate per Gyr was set to 10 times the initialgas content of the region. As the inflow is enriched, thislarge value ensures that the number of low metallicity stars

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 11

Figure 6. Hβ vs Fe4668 for S0 galaxies (Fisher et al. 1996) and the 4 spiral bulges presented here. (N.B. Gorgas et al. (1997) did notinclude Fe4668.) Dotted lines indicate Worthey’s SSP values for metallicities; -0.25 0.0, +0.25, +0.5 and ages of 1.5 to 17 Gyr. Solid anddashed lines are as in fig. 4 (see section 4).

produced is kept small. Significantly higher values of inflowrate had negligible effects on the achieved line-strengths. Tobe consistent with dynamical models of primordial collapsethe period of inflow was stopped after 0.4 Gyr while the SFwas allowed to continue to the present day, albeit at an everdecreasing rate as gas is consumed. This generates an ∼ 1Gyr burst of SF, which is also consistent with the chemo-dynamical models of Theis et al. (1992). Continuation of SFto the present day also maximises the predicted strengthsof metallicity dependent lines. To demonstrate the range ofmetallicities such models can achieve, star formation effi-ciency (C) during the inflow period was varied between 0.2and 4.0 Gyr−1. This range of SF efficiencies spans the val-ues given by Fritz-v. Alvensleben & Gerhard (1996) for Sb,Sa and E galaxies up to values for ultra-luminous star-burstgalaxies. After the initial 0.4 Gyr of inflow, SF was allowedto continue in all models with an efficiency C=0.2. The agesof the models varied between 1.5 and 17 Gyr. The results ofour GCE models of primordial collapse for <Fe>, Mg2, Hβ

and Fe4668 are shown as solid lines in figs. 4 to 6. Compari-son of indices for E, S0 and spiral bulges (fig. 4) to the rangeof values achieved by primordial collapse models shows theinability of these models to reproduce the observed centralline-strengths for both <Fe> and Mg2 features. This resultis confirmed in figs 5 and 6. Introduction to models of a bi-ased IMF at early epochs can resolve the under-abundance

problem but can not simultaneously explain the high Hβ

values observed in some elliptical galaxies and at least oneof our spiral bulges (fig. 5). We might also expect to see anMg2 excess with respect to solar in the biased IMF scenario.There is no evidence for this excess in our bulge data.

4.2 Merger Models.

While direct evidence for primordial collapse is not observed,examples of ongoing galaxy mergers are. Indeed, Schweizer& Seitzer (1992) suggest that, based on the number of galax-ies exhibiting on-going mergers or the fine structure indica-tive of a recent event, ∼ 50% of field ellipticals have under-gone a merger event in the last 7 Gyr. This would seem tobe supported by the fact that ∼ 50% of the elliptical and S0sample shown in fig. 5 show Hβ values that lie above the 8Gyr SSP line. In merger models, galaxy formation proceedsby coalescence of fragments that have undergone SF prior tomerger. Such models can achieve high central line-strengthsas the period of SF prior to merger pre-enriches the ISMfrom which the bulk of the population is formed (Sansom &Proctor 1998). It should be noted that both primordial col-lapse and merger models of galaxy formation predict shortbursts of SF due to the extremely rapid inflow of gas inboth scenarios (Theis et al. 1992; Barnes & Hernquist 1996).

c© RAS, MNRAS 000, 1–14

12 R. Proctor et al.

Figure 7. Index against log radius (radius in arcsec) for <Fe>, Fe4668, Mg2 and Hβ indices of the 4 galaxies of the Palomar data set.Open squares are disc contaminated side of the galaxies. All data points within the regions spanned by the regression lines are includedin the fit. Note the flattening in metallicity sensitive indices in NGC 3623 which may be associated with the presence of a bar.

Therefore, the excess Mg observed in elliptical galaxies ispredicted by both of these models, without the need for theassumption of a biased IMF. The modelling of spiral bulges,using merger models, is constrained by their lack of Mg2 ex-cess. As the time-scale of gas inflow during an interaction isextremely rapid, large merger induced star-bursts will resultin an Mg2 excess. As this is not seen in our spiral bulges, ifbulge formation is dominated by accretions/mergers, the SFbursts must have been small and numerous. This is consis-tent with the presence of the disc which is highly unstableto mergers of more than a few percent of the total galaxymass (Toth & Ostriker 1992). While we can construct singleburst merger models that reproduce the <Fe>, Hβ and Mg2indices observed in the bulges, detailed modelling of mergerscenarios has not been carried out as the large parameterspace implied by multiple short bursts makes these modelshighly degenerate. In the limiting case, where interactionsare so frequent that the induced SF is effectively continu-ous, we expect merger models converge to extended inflowmodels.

4.3 Models with Extended Inflow.

Extended inflow models represent long term inflow of gasinto the region being modelled (e.g. Samland et al. 1997).The long duration of the inflow is consistent with formationof the bulge both by in-fall of gas from the disc (perhaps bythe action of bars) as well as by the accretion of gas from thehalo or captured satellite galaxies over an extended period.

The models assume constant SF efficiency (C) with rela-tively modest gas inflow over a period of 2 to 17 Gyr andmodel spectral indices at a variety of points in the historyof the population. With reasonable choices for C (0.4 - 4.0Gyr−1) and inflow rate (105- 106 M⊙ Gyr−1) the observed<Fe>, Mg2 and Hβ in bulges can simultaneously be repro-duced to within errors. An inflow rate of 106 M⊙ Gyr−1

corresponds to an increase in mass in the volume modelledequal to the initial mass every Gyr. The range of SF effi-ciencies excludes low values (0<C<0.4) as, even with con-tinuous SF for 17 Gyr, there is insufficient star formation toproduce the metals required to achieve the high central line-strengths of NGC 3623 and NGC 4565. Formation episodesof <∼ 2 Gyr result in an Mg2 enhancement not reflected inthe data (using the SSP models of Weiss et al. 1995). Figs.4 to 6 show the range of indices achieved by a selcetion ofextended inflow models that reproduce the indices of thespiral bulges in both the Mg2 vs <Fe> and <Fe> vs Hβ

planes to within the errors. However, it should be notedthat the models are highly degenerate with respect to in-flow rate and duration, SF efficiency, and time of SF onset(galaxy age). Consequently, the models are only shown toillustrate their ability to achieve the indices. The range ofSF efficiencies and inflow durations that can reproduce the<Fe>, Mg2 and Hβ indices in our 4 bulges are consistentwith the findings of Samland, Hensler & Theis (1997) usinga chemo-dynamical model of disc galaxy formation. Blumet al. (1996) showed that red giants in the Galactic bulgepossess approximately solar metallicities and suggest thatmultiple epochs of SF have occurred in the centre of the

c© RAS, MNRAS 000, 1–14

Constraining the Star Formation Histories of Spiral Bulges. 13

bulge in the last 7 - 100 Myr. Consequently, extended inflowmodels of spiral bulge formation are, so far, entirely consis-tent with both observations of Lick indices, the stars in ourown bulge and chemo-dynamical modelling.

5 CONCLUSIONS.

Our study of the bulges of 4 spiral galaxies using theLick system of spectral indices has shown that central line-strengths are high. <Fe> indices are similar to those foundin the centres of elliptical galaxies. However, a difference inMg2 index at a given <Fe> exists between elliptical galaxiesand our sample of spiral bulges. Spiral bulges lie within theregion of the Mg2 vs<Fe> plane occupied by the solar abun-dance ratio SSPs of Worthey (1994), while ellipticals exhibitenhanced Mg2 (the known [Mg/Fe] excess). This differencebetween the two object types reflects differences in their starformation histories. Our data are consistent with both cor-relations of Mg2 with central velocity dispersion and Mg2gradient with central velocity dispersion observed in ellipti-cal galaxies.

Using our GCE code, we have shown that the centralline-strengths in spiral bulges cannot be achieved with pri-mordial collapse models of spheroid formation with or with-out the assumption of constant, Salpeter IMF. Indeed, theinferred solar [Mg/Fe] argues against a biased IMF in spiralbulges. Models of bulge formation with gas inflow and starformation extended over 2-17 Gyr can achieve the observedcentral line-strengths in all of our sample. These models areconsistent with the chemo-dynamical modelling of Samland,Hensler & Theis (1997) as well as observations of individualstars in the bulge of our own Galaxy (Minniti et al. 1992;Blum et al. 1996). It has been shown that at least one bulge(NGC 5689) must contain a population of relatively youngstars (<∼ 5 Gyr) again consistent with extended inflow mod-els of bulge formation.

NGC 3623 shows the presence of kinematic substruc-ture. A dip in velocity dispersion is observed in the centreof this galaxy coincident with a sharp peak in metallicitysensitive indices. This structure suggests the presence of abar, or perhaps disc, at the centre of this galaxy. Analysis ofa much larger sample of galaxies, including a repeat obser-vation of NGC 3623, is being carried out with observationsfrom the WHT, to test the findings of this paper. Thesedata will be presented in a future paper and will includeother age sensitive indices, such as the newly calibrated Hδ

and Hγ indices, which are less affected by emission than theHβ index. Increased accuracy of age determination, coupledwith the larger number of metallicity sensitive indices, willpermit tighter constraint of our models of bulge formationand their star formation histories.

Acknowledgments.

We thank our referee D. Friedli for his constructive com-ments. The authors acknowledge the data analysis facilitiesprovide by the Starlink Project which is run by CCLRCon behalf of PPARC. In addition, the IRAF software pack-age was used. IRAF is distributed by the National Opti-cal Astronomy Observatories, which is operated by AURA,Inc., under cooperative agreement with the National ScienceFoundation. This work is based on observations obtained at

Palomar observatory, which is owned and operated by theCalifornia Institute of Technology.

REFERENCES

Balcells M., Peletier R.F., 1994, AJ, 107,135Barnes J.E., Hernquist L., 1996, ApJ, 471,115Bender R., Burstein D., Faber S.M., 1993, ApJ, 411,153Bertola F., Cinzano P., Corsini E.M., Pizzella A., Persic M.,

Salucci P., ApJ, 458,L67Blum R.D., Sellgren K., Depoy D.L., 1996, AJ, 112,1988Bottema R., Gerritsen J.P.E., 1997 MNRAS, 290,585Burkhead M.S., Hutter D.J., AJ, 86,523Carlberg R.G., 1984, ApJ, 286,403Carollo C.M., Danziger I.J., Buson L., 1993, MNRAS, 265,553Davies R.L., Sadler E.M. Peletier R.F, 1993, MNRAS, 262,650de Vaucouleurs G., de Vaucouleurs G., Corwin H.G., Buta R.J.,

Fouque P., Paturel G., 1991, Third Reference Catalogue ofBright Galaxies. Springer-Verlag, New York

Faber S.M., Friel E.D., Burstein D., Gaskell C.M., 1985, ApJS,57,711

Fisher D., Franx M., Illingworth G., 1995, ApJ, 448,119Fisher D., Franx M., Illingworth G., 1996, ApJ, 459,110Fritze-v. Alvensleben U., Gerhard O.E., 1994, A&A, 285,751Friedli D., Benz W., Kennicutt R., 1994, ApJ, 430,L105

Friedli D., 1996, A&A, 312,761Friedli D., 1998, in Friedli D., Edmunds M., Robert C., Drissen

L., eds., ASP conf. Ser., Vol. 147, Abundance Profiles: Di-agnostics Tools for Galaxy History. Astron. Soc. Pac., SanFrancisco, p. 287

Frogel J.A., 1998, Galaxy Evolution: Connecting the Distant Uni-verse with the Local Fossil Record, Observatoire de Paris-Meudon, in press (astro-ph/9812034).

Gibson B.K., Matteucci F., 1997, MNRAS, 291,L8Gorgas J., Faber S.M., Burstein D., Gonzalez J.J., Courteau S.,

Prosser C. 1993, ApJS, 86,153Gorgas J., Pedraz S., Guzman R., Cardiel N., Gonzalez J.J., 1997,

ApJ, 481,L19Goudfrooij P., Emsellem E., 1996, A&A, 306,L45Greggio L., 1997, MNRAS, 285,151Gustafsson B., Karlsson T., Olsson E., Edvardsson B., Ryde N.,

1999, A&A, 342, 426Guthrie B.N.G., 1992, A&AS, 93,255Hes R., Peletier R.F., 1993, A&A, 268,539Jablonka P., Martin P., Arimoto N., 1996, AJ, 112,1415

Jones L., 1996, PhD thesis, Univ. North CarolinaKeel W.C., 1983, ApJ, 466,486Kennicutt R.C., 1989, ApJ, 344,685Maeder A., 1992, A&A, 264,105Minniti D., Olszewski E., Liebert J., White S.D.M., Hill J.M.,

Irwin M.J., 1995, MNRAS, 277,1293Nomoto K., Thielmann F.K., Yokoi K., 1984, ApJ, 286,644Oke J.B., Gunn J.E., 1982, PASP, 94,586Renzini A., Voli M., 1981, A&A, 94,175Rice W., Merrill K.M., Gatley I., Gillett F.C., 1996 AJ, 112,114Samland M., Hensler G., Theis Ch., 1997, ApJ, 476,544Sansom A.E., Peace G., Dodd M., 1994, MNRAS, 271,39Sansom A.E., Proctor R.N., 1998, MNRAS, 297,953Sansom A.E., Proctor R.N., Reid I.N. 1998, in Friedli D., Ed-

munds M., Robert C., Drissen L., eds., ASP conf. Ser., Vol.147, Abundance Profiles: Diagnostics Tools for Galaxy His-

tory. Astron. Soc. Pac., San Francisco, p. 26Sil’chenko O.K., 1993, Pis’ma Astrom. Zh, 19,701Schweizer F., Seitzer P., 1992, AJ, 104,1039Theis C., Burkert A., & Hensler G., 1992, A&A, 265,465Toth G., Ostriker J.P., 1992, A&A, 265,465Tripicco M.J., Bell R.A., 1995, AJ, 110,3035

c© RAS, MNRAS 000, 1–14

14 R. Proctor et al.

Tully R.B., 1988, Nearby Galaxies Catalogue, Camb. Univ. Press,

CambridgeVazdekis A., Peletier R.F., Beckman J.E., Casuso E., 1997, ApJS,

111,203Weiss A., Peletier R.F., Matteucci F., 1995, A&A, 296,73Woosley S.E., Weaver T.A., 1995, ApJS, 101,181Worthey G., 1994, ApJS, 95,107Worthey G., 1998, PASP, 110,888Worthey G., Dorman B., Jones L.A., 1996, AJ, 112,948Worthey G., Faber S.M., Gonzalez, J.J., 1992, ApJ, 398,69Worthey G., Faber S.M., Gonzalez, J.J., Burnstein D., 1994,

ApJS, 94,687Worthey G., Ottaviani D.L., 1997, ApJS, 111,377

c© RAS, MNRAS 000, 1–14